The specimens studied here were generally identical to Schmeissneria , especially the holotype of S. microstachys (pl. I, fig. 1) . The latter closely resembled S. sinensis in spicate structure, paired female organs, female organ morphology, longitudinally ribbed axes of female structures, and the insertion of a female structure on the apex of a short shoot, even though the specimens described here and S. microstachys were from different continents and geological epochs, and S. sinensis had more densely clustered female organs and much shorter peduncles of female organ pairs (0.5 mm rather than 2 mm long). Fortunately, the internal structure was preserved in S. sinensis and thus shed new light on its affinity.
Structure and Interpretation
In all non-angiospermous seed plants, there is an opening at the nucellar apices to allow the pollen grains to approach the ovules [31,32]. The dimension of the opening has to be large enough to let pollen grains pass through. Unlike seed ferns or other gymnosperms, no opening was found at the apex of the central unit in S. sinensis (Figs. 3d–f). In the view from the interior of the central unit to its apex (Fig. 3f), any opening larger than 20 μm (the normal size of an average pollen grain) should be visible if present on the central unit wall. Although the preservation of the fossils reported here was not perfect, the female organs had both the cellular details (Fig. 3b) and a septum only about 10 μm thick (Fig. 3i) preserved. The preservation of such fine details suggested that preservation fidelity was high enough for structures above the cellular level. Therefore the central unit apex of S. sinensis was physically closed, at least to pollen grains. This was essentially different from the situation in Caytonia, in which there are numerous ovules within a cupule that has an opening for pollen grain entry before pollination [33,34]. There were two alternative interpretations for the absence of pollen entry in Schmeissneria: one was that the pollen entry did not exist at all; the other was that the pollen entry had been obliterated. The latter situation has been seen in Gnetum [32,35], Ephedra, Pinus, Cedrus, Cephalotaxus , and possibly in Caytonia [33,34], in all of which the pollen canals were plugged or obliterated by tissue outgrowth or cell proliferation after pollination. This tissue growth or cell proliferation is concomitant with morphological changes [35,36]. However, in the case of Schmeissneria this alternative appeared unlikely for the following reasons: 1) the female organs in Figs. 3d–f were in their early stages (pre-pollination), suggested by their smaller sizes and morphology different from winged seeds ; 2) there was neither any trace of abnormal tissue outgrowth nor sudden changes in the wall at the apex of the central unit (Figs. 3d–f), unlike what is observed in Gnetum , Cephalotaxus, and Ephedra . Consequently, the other alternative was more reasonable and acceptable. This was the major reason that the authors correlated the structures of Schmeissneria's central unit, envelope, female organ and female structure with an angiosperm's gynoecium, perianth, flower and inflorescence.
A septum extended from the base (Figs. 2g, 4b), through the middle (Figs. 3g–h), to the apex (Figs. 2h, 3e,f,i, 4c) of the central unit. The septum separated the central unit into two independent locules (Figs. 2f–h, 3h–i, 4b–c). The latter were suggestive of two carpels in a central unit (equivalent to gynoecium in angiosperm) . The septum appeared as a ridge raised above the smooth wall of the central unit in the longitudinal view (Fig. 2g). Its smooth connections to the side walls (Figs. 3h–i) and base (Fig. 2g) of the central units suggested that the septum represented an original structure rather than an artefact or alteration. The repeated presence of a septum of various poses in four individual female organs (Figs. 2g–h, 3d–i) also strongly suggested its truthful existence. Up until now there has been no report of a complete septum in a seed or ovule in any gymnosperm. Sometimes paired ovules in Ginkgo may appear to have a membraneous division in between. However, their ovules have never been completely enclosed before pollination, the tips of the paired ovules point to different directions, and there is no empty interior space within any ovule. Therefore the paired ovules of Ginkgo were distinct from the central unit of Schmeissneria.
Schmeissneria sinensis had female organs of various ontogenetic stages, from small premature ones at the top to large mature ones at the base of the female structures (Figs. 1, 2, 3, 4). Considering the slight morphological difference between the topmost (most immature) and bottommost (most mature) female organs, it was reasonable to assume that the female organs reported here, especially the smaller ones, were not yet pollinated. The rough internal surface in the upper portion of the central unit (Fig. 2g and Fig. 4b(g)) and the longitudinal ribs on it (Fig. 3e), in contrast to the smooth surface in the lower portion of the internal surface of the central unit (Fig. 2g and Fig. 4b(a,c)), suggested that the upper portion of the central unit was empty, while the lower part was occupied by another inherent substructure (probably an ovule). This situation was unlikely in ovules or seeds, which rarely, if ever, have an empty interior space. It was also unlikely to be a result of differentiated preservation related to histological differences because the presence or absence of tissues is a morphological rather than histological character. The winged seeds connected to vegetative parts reported by Kirchner and Van Konijnenburg-Van Cittert (Fig. 1b, ) were distinct from the female organs reported here, implying the immaturity of S. sinensis. Considering the aforementioned details, since Wcislo-Luraniec  and Kirchner and Van Konijnenburg-Van Cittert  have proven the female nature of Schmeissneria, the immature central unit with two locules and a closed apex sheathed by an envelope can be interpreted as a gynoecium of two carpels surrounded by a perianth. This suggested that the carpels (gynoecium) in Schmeissneria were closed before pollination, a situation quite different from the "angiospermy" after pollination in gymnosperms .
In 1838, Presl (pl. 33, fig. 12 only) reported fructifications from Keuper Sandstone (actually Liassic age ) of Reundorf near Bamberg, Germany . He considered them male flowers of Pinites microstachys (Conifers). Schenk (pl. XLIV, figs. 11, 12 only, 1867) investigated similar fossils from Veitlahm near Kulmbach, describing them as female flowers of Stachyopitys preslii, which was associated with Schizolepis (a coniferous genus) . Later Schenk (fig. 180b only, 1890) assigned Stachyopitys preslii as male fructifications of Baiera münsteriana (Ginkgoales), and interpreted the formerly "female fructifications" as male flowers in an early stage . This association between Baiera and Stachyopitys proposed by Schenk  was later widely accepted (taf. 29, abb. 4 ; fig. 239c ; fig. 375d ; abb. 303h ; fig. 198B ; taf. 1, fig. lower left ; fig. 35d ). Actually, fossils of various affinities had been lumped into Stachyopitys preslii . Wcislo-Luraniec (pl. 1 and textfigs. 1, 2, 1992) cast doubt on the male nature of Stachyopitys preslii because of the discovery of "cupules," thereby proving the female nature of the fossil . At the same time, Stachyopitys preslii was found connected with Weber's (p58, taf. 5, fig. 51 ) Glossophyllum? sp. A . With more complete fossil materials available, a new genus, Schmeissneria (Ginkgoales), was established for the female Stachyopitys and officially separated from the male Stachyopitys (Figure 1 and pls. I-III, pl. IV, fig. 1, ). It appeared that Kirchner and Van Konijnenburg-Van Cittert did not realize that they had undermined the ground for the ginkgoalean affinity of the fossil and uncritically accepted the previous conclusion as true . Furthermore, they did not give any valid reason for why they put the genus in Ginkgoales except for comparing the paired "ovules" of Schmeissneria and the paired ovules of Ginkgo in two sentences (p.207) . Apparently, their comparison was not sufficient to place the fossil in Ginkgoales with confidence because the characteristic collar at the base of Ginkgo seed was absent in all specimens of Schmeissneria and the winged seeds of Schmeissneria were never seen in any Ginkgoales. However, re-examining the references indicated that 1) the discovery of the connection between Stachyopitys (Schmeissneria) preslii and Weber's Glossophyllum? sp. A disproved the relationship between Stachyopitys (Schmeissneria) preslii and Baiera münsteriana proposed by Schenk ; and 2) Glossophyllum? sp. A did not belong to the ginkgoalean Glossophyllum . Therefore, Schmeissneria was dangling phylogenetically. Since the connected vegetative organs could not resolve its affinity, the affinity of Schmeissneria had to be resolved based on the internal structure of its own female reproductive organs.
Seed plants include two major groups, gymnosperms and angiosperms. Currently the known major Mesozoic and extant gymnospermous groups include Ginkgoales, Cycadales, Bennettitales, Coniferales, Glossopteridales, Caytoniales, and Gnetales. Among them, Cycadales have pinnate leaves and ovules/seeds attached to the margins of megasporophylls [5,31,32]; Bennettitales have pinnate leaves and conspicuous cone-like reproductive organs with ovules/seeds arranged on a central receptacle [5,31,32]; Coniferales have needle-like or various leaves and more or less cone-like structures composed of bract-scale complexes, except for all or some elements of Taxaceae and Podocarpaceae [5,31,32,45], the latter two are distinctly different from Schmeissneria; Glossopteridales have reticulate leaves and their megasporophylls adnated to the adaxial of the foliage [5,31,32,46]; Caytoniales have reticulate leaves and cupules with inverted openings oppositely arranged along the laterals of the megasporophyll axis [5,32]; and Gnetales have a micropylar tube, articulate shoot, opposite/whorled leaves, opposite/whorled bracts in the reproductive cone, ovuliferous units in the bract axil [31,32,47,48] (Table 1). These characteristics clearly distinguished these groups from Schmeissneria. Therefore these groups will not be considered further in the following discussion on the affinity of Schmeissneria, and the only possibilities remaining are Ginkgoales, Angiosperms, or a new group of seed plants.
Several characters are used to distinguish angiosperms and gymnosperms, including enclosed ovules/seeds, double fertilization, vessel elements, reticulate venation, and tectate-columellate pollen wall structure. However, it is harder to draw a line between these two groups than it might appear: neither of the above characters is a touchstone for angiosperms [1,3]. None of the above characters is unique to angiosperms. All the ovules in gymnosperms are exposed when pollination occurs, but at least some of the ovules/seeds are enclosed after fertilization (Caytoniales ; Gnetales ; Coniferales and Gnetales ; Coniferales ). Angiosperms have completely closed carpels. However, in some basal groups this closure is by secretion, and not by postgenital fusion (e.g., Amborellaceae, Schisandraceae, Austrobaileyaceae, Trimeniaceae ). Double fertilization has been reported in non-angiosperms, as in Ephedra [31,50] and Abies . Vessel elements are also present in various non-angiosperms, such as Selaginella, Equisetum, Pteridium , Gigantopteriales , and Gnetales [1,4,31,32]. Reticulate venation has been reported in non-angiosperms, including Dipteridaceae [52-55], Gigantopteriales [56,57], Caytoniales [33,58], Glossopteridales [5,40,46,58], Bennettitales , and Gnetales [31,32]. Pollen grains with an angiosperm-like wall structure have been reported in many pre-Cretaceous plants [59-61] that are regarded as non-angiosperms by others . According to Tomlinson and Takaso , the only consistent difference between angiosperms and gymnosperms is that the ovules at pollination are exposed in gymnosperms, but enclosed in angiosperms. Fortunately, the closed carpel at or before pollination is a character that is sufficient to identify an angiosperm alone. This was one of the characters used here to resolve the affinity of Schmeissneria.
Two characters separated Schmeissneria from known gymnosperms: the vertical complete septum and the closed apex of the central unit. Considering all available information and the definitions of plant groups, there were two alternatives left to us: 1) accepting that Schmeissneria as a new angiosperm in the Jurassic, or 2) proposing Schmeissneria as a new gymnosperm. Although the specialized features of the early Cretaceous angiosperm Archaefructus  and other data [8,9] may imply the possible existence of angiosperms before the Cretaceous, Schmeissneria did not look like any known typical angiosperm. However, this dissimilarity was conceivable and understandable since 1) "angiophytes" had not evolved any typical identifiable angiospermous character , and 2) many angiospermous taxa were much more diversified then and much of that diversification had since become extinct . Extreme caution should be exercised when a Jurassic angiosperm, along with their relatives of which we know very little, is compared with the extant angiosperms.
If accepted as an angiosperm, because of its early Jurassic age in Europe, Schmeissneria would push the origin of angiospermy back to the Triassic. This would make the claims of Triassic angiosperms [13,59-61], [63,64] less surprising, and also help to bridge the gap between the fossil record [3,5,10-12,65] and molecular data [66-72]. However, it should be kept in mind that Schmeissneria might well represent early angiosperms still sporadic in the vegetation dominated by gymnosperms, that it might bear no direct relationship with any known angiosperm, and that the presence of Schmeissneria in the flora was still far different from the radiation and diversification of angiosperms.